To provide High Availability for applications is a standard requirement for todays data centres. High Availability ist often achieved with an exceptional increase of complexity like additional Cluster Interconnect (heartbeat) or a Quorum device. With OSL Storage Cluster there is no need for any heartbeet nor Quorum device.
Numerous data-center applications (databases) must be online without interruption for economic reasons. To meet this requirement, effective protection against hardware failures is inevitable. State of the art are clustered solutions. They care about a migration of the applications to another piece of hardware in the case of disasters or failures. A ready for failover host takes over the application. Off-times are avoided. This approach assumes, however, an abstract access to hardware resources. Processes and applications have to be described and adjusted if necessary.
In the field of conventional clustering active/active variants are the best known ones. There at least two application servers are online and take over the tasks when another fails. Furthermore, active/passive solutions are established: A standby unit jumps in and boots if the primary system fails. While such architectures are also supported, the cluster concept of OSL is not restricetd to these two traditional variants. With OSL Storage Cluster much more intelligent configurations can be built. The OSL-user assigns priorities to her applications. On the basis of performance benchmarks, application descriptions, and these priorities an assignment of applications to different servers of the cluster is done (1). This can be automated according to the performance of available hardware (2). If a cluster node fails, a redistribution of application on the still available cluster nodes occurs on the basis of required hardware resources. Higher priority applications can thereby displace even less priority ones, in order to remain online (3). Then the user may prioritize productive applications higher and test, development or QA applications even lower for being best protected against hardware failures or even disaster scenarios. Later, if the failed hardware is back online, the stopped applications can get restarted (4). A further migration of running applications is not necessary anymore.
Integration of Applications with Virtualized Runtime Environments
Based on the described technology OSL Cluster Storage Cluster provides the required system resources via standardized, virtualized interfaces to your applications. Because of adaptations of environment parameters of applications this concept is referred to as virtual runtime environments. Compared to the widely used virtual machine, virtual runtime environments are primarily distinguished by easy management based on a clear organization of processes and services. Because hardware is not abstracted with a complex hypervisor, applications work in virtualized runtime environments with the performance of a native execution.
The Application as a Focal Point
Running applications are the most valuable commodity in data centers. As long as your applications keep running without losing data, one usually gets over hardware failures. The virtual runtime environments are putting applications consistently in the center of data center operations. Only real application awareness helps to face other challenges such as backup, archiving and disaster recovery, system copies, or even storage management convincingly. According to the concepts of OSL processes are designed by intelligent software from the perspective of the application in an optimal way.